Exploring the rhetorical semiotic brand image structure of ad films with multivariate mapping techniques

DOI 10.1515/sem-2014-0012 Semiotica 2014; 200: 335 – 358
George Rossolatos
Exploring the rhetorical semiotic brand
image structure of ad films with multivariate
mapping techniques
Abstract: The aim of this paper is to demonstrate the applicability of multivariate
mapping techniques to the exploration of the rhetorical semiotic brand image
structure of ad films. By drawing on correspondence analysis and multidimen
sional scaling, two techniques that are amply used in corpus linguistics and in
marketing research, but also on the data reduction technique of factor analysis, it
will be displayed how a set of nuclear semes and classemes or an intended semic
structure that underlies ad filmic discursive structures may be projected along
side rhetorical figures by its internal stakeholders (i.e., a brand management
team, an account planning team or a marketing research team) with view to
attaining differential brand associations. The illustration of the exploratory ana
lytical methods takes place by recourse to a corpus of 62 ad filmic texts from 13
sub-brands of the 3 major brands in the UK cereals market and 321 ad filmic seg
ments that resulted from the segmentation procedure. This paper seeks to con
tribute to advancements in the brand image and advertising rhetoric research
streams by addressing distinctive modes of advertising rhetorical configuration at
the level of the ad filmic text (as against print ads where the bulk of research on
advertising rhetoric has concentrated), and moreover on a segment-by-segment
level, rather than treating the ad film as a standalone unit of analysis, by adopt
ing a multimodal outlook to advertising configurations that takes into account
not only the verbal or the visual mode, but also interactions among modes, by
adopting a product category-specific approach to advertising/brand textuality
that draws on rhetorical semiotics and by employing exploratory statistical
methods against the background of content analytic output.
Keywords: rhetorical semiotics; ad films; brand image; multivariate mapping
techniques
George Rossolatos: University of Kassel. E-mail: georgerossolatos123@gmail.com
Authenticated | georgerossolatos123@gmail.com author's copy
Download Date | 5/23/14 7:10 AM

336 George Rossolatos
1 Introduction
The meaning and shaping forces of brand image have been extensively researched
in the marketing literature (Dobni and Zinkhan 1990; Stern 2001). “Brand image
can be defined as perceptions about a brand as reflected by the brand associa
tions held in consumer memory” (Keller 1998: 93). Brand associations contain the
meaning of a brand, hence their value is instrumental as determinants of brand
meaning. Brand associations may be further split into core and peripheral associ
ations or associations that are instrumental for building and maintaining a
brand’s core (DNA) and associations that are responsible for enriching a brand’s
semantic nucleus. It is particularly important to distinguish between these two
types of brand image attributes insofar as they circumscribe the essential and
non-essential elements of a brand knowledge structure and hence impact directly
on a brand’s intended positioning, as well as the extent to which an intended
positioning is in fact reflected in consumers’ perceptions.
Advertising constitutes one of the principal semiotic modes whereby brand
meaning is generated. “Traditionally, advertising has been a particularly power
ful way of communicating a brand’s functional values, as well as building and
communicating its emotional values” (De Chernatony 2006: 6). “Advertising has
become such a pervasive mode of semiosis in today’s advanced economies that it
is now an essential way of knowing the world, particularly through which the
arbitrary and culturally determined are made to seem necessary and natural,
even as a society is constantly evolving” (Mick et al. 2004: 26).
The vast majority of analyses that have been offered in this research stream
pertain to the “decoding” side of advertising and concomitantly to the already
attained transformation of advertising expressive elements into brand image at
tributes, with an undue emphasis on the very encoding process of brand texts
(see Solomon and Greenberg 1993). Scholars in the advertising-related literature
(e.g., Goldenberg and Mazursky 2008) have made attempts at discerning “depth
structures” of ad expressive elements, albeit against a non-rhetorically semiotic
informed conceptual background, while linking such “depth structures” neither
to intended brand signification, nor to a brand’s semic nucleus as the essential
correlate at the plane of content and in a product category-specific framework
(Rossolatos forthcoming).
The main argument that is put forward in this paper is that unless a brand
image structure is projected in the first place in such a manner that adjoins an
intended semic universe to a selected expressive inventory, it is impossible to
gauge and furthermore to manage to what extent the resulting brand associations
in consumers’ minds do in fact derive from a brand’s communicated expressive
inventory (Rossolatos 2014). To this end, the ensuing structuralist rhetorical

Rhetorical semiotic brand image structure 337
semiotic framework for projecting an intended brand image structure aims at
bridging this gap between what is intended and what is perceived by the final
consumer, with the employment of content analysis and the application of multi
variate mapping techniques and factor analysis to the content analytic output.
2 Rhetorical research in advertising: A heavily
fragmented research field
Various perspectives, typologies, and taxonomies spanning different disciplines
have been offered in order to account for the modes of rhetorical configuration of
both static and moving images, such as Foss’s visual rhetoric (2005), Groupe μ’s
(1992) Traité du Signe Visuel (‘Treatise on Visual Rhetoric’) from a rhetorical semi
otic point of view, Kostelnick and Roberts’ (2010) from a visual design point of
view, McQuarrie and Mick’s (1996), McQuarrie and Phillips’s (2004), McQuarrie’s
(2008) from a consumer research point of view.
Groupe μ (1992) provided an updated account of the rhetorical operations
(adjunction, suppression, substitution, permutation)1
that were featured in their
first rhetorical treatise (1970), while applying them to visual signs that had been
introduced in the first treatise (1970), with further qualifications (see Rossolatos
2012)2
.
The rhetorical perspective that is offered in this paper has been edified on
the taxonomy of rhetorical figures offered by Groupe μ in their first rhetoric (see
Rossolatos 2013c), while taking into account for analytical (and not taxonomic)
purposes how rhetorical operations function in the province of visual signs. Fur
thermore, the definitional and operational scope of the involved operations and
figures was expanded in order to address the particularities of ad films (Rossola
tos 2013c).
Efforts have been undertaken at extending the application of traditional fig
ures mostly to static images (e.g., visual metonymy, Willerton 2005; visual hyper
bole, Callister and Stern 2007) and print advertisements (e.g., Durand 1970, 1987;
Tom and Eves 1999; McQuarrie and Phillips 2004). The extension of figures to
moving images and advertising filmic narratives has been mostly limited to visual
1 These key operations stem from a long tradition, starting with Quintilian (see Sloane 2001 and
Nöth 1990).
2 A similar strategy of retaining operations, but dropping figures was pursued by McQuarrie and
Philips (2004) in their taxonomy of operations in print ads.

metaphors and visual metonymies (e.g., Forceville 2007; Forceville and Urios-
Aparisi 2009; also see Callister and Stern 2007: 3). By adopting a common classi
fication framework for visual and verbal elements, both purely verbal, purely
visual, but also verbo-visual figures and operations may be discerned as being
operative in ad filmic texts, while taking into account the evocative powers of the
visual mode, namely, “immediacy, verisimilitude, and concreteness that help in
fluence acceptance in ways not available to the verbal” (Foss 2004: 314). To this
end, the rhetorical figures were redefined in order to encapsulate visual com
ponents, but also modes of interaction between visual and verbal modes in ad
filmic texts, as laid out in Rossolatos (2013c). These figures aim to encapsulate
salient and frequently recurring modes of rhetorical configuration of ad filmic
texts.
3 Rhetorical semiotic conceptual framework for
projecting a brand image structure
The conceptual panoply that is employed in this study draws on structuralist
rhetorical semiotic terminology, and particularly on the semiotic theory of A. J.
Greimas (1966, 1970, 1971, 1976a, 1976b, 1983, 1987; also see Mick 1986, 1987) and
the rhetorical semiotic approach to visual and verbal signification that was pro
pounded by Groupe μ (1970, 1992; Rossolatos 2013a, 2013b). The vantage point of
Greimas’s epistemological edifice rests with a quest for the primary elements of
signification. Greimas’s method, which appeared initially in the form of a seman
tic theory (1966) and was further elaborated into a full-fledged textual semiotic
conceptual armory and methodology, concerns primarily relations and transfor
mations. Brand meaning arises only through relations among elements from the
planes of expression and content, while it takes place through various transfor
mations in different levels or strata [niveaus] of the so-called generative trajectory
of signification (Greimas 1970; Rossolatos 2012). Structuralist semiotics and par
ticularly a generativist approach to the generation and management of brand
meaning over time constitutes a most pertinent blueprint (Rossolatos 2013a).
The analytical route that is proposed in this paper seeks to demonstrate how
a brand’s semic or brand image structure made up of elementary semantic units
(semes) may be explored in relation to an ad text’s modes of rhetorical configura
tion as rhetorical figures. Rhetorical figures have been shown by Greimas to be
indispensable relata that cut across an entire trajectory of signification (spanning
an elementary structure of signification, an intermediate structure at a semio-

narrative level and a manifest or discursive structure; cf. Rossolatos 2012, 2014,
2013e), while acting as the “semiotic glue” that adjoins morphologically distinc
tive elements from the three strata. Furthermore, the proposed analysis of ad
filmic texts pursues a combined synchronic/diachronic approach, in line with
studies in semiotics, but also in corpus linguistics that are geared towards an
understanding of the historical evolution of language (see, for example, Nöth
1990: 63; Wichmann 2008: 194), by attending to ad filmic texts’ structuration in
the context of both single filmic texts and across ad films. “Synchronically, we see
the set of interpretive conventions as a way to understand the contemporary
‘meaning’ of a brand. By interpreting the evolution of . . . brands, we see the way
the . . . meaning of these brands alters over time (i.e., diachronically)” (Hatch and
Rubin 2006: 57). In this paper, the findings from the consolidated, diachronic
analysis are presented (both levels are addressed more extensively in Rossolatos
forthcoming).
The study draws on Groupe μ’s typology of rhetorical figures that was put
forward in their first rhetorical treatise, while expanding the definitions of these
figures in order to accommodate the visual modality that is prominent in ad films.
I shall refrain from citing anew the list of figures that were employed during the
coding procedure of ad filmic segments, which may be found in Rossolatos 2013d,
2013f.
4 Methodology for projecting a rhetorical
semiotic brand image structure
The first step in this exploratory endeavor consists of segmenting the selected
corpus of ad films in individual segments with the employment of the content
analytic software atlas.ti (see Rossolatos 2013c, 2013d, 2013f for further details on
the segmentation procedure). The segmentation procedure of the 62 ad films from
13 sub-brands of the 3 key brand players that make up the selected corpus (the UK
cereals market) resulted in 321 filmic segments (further details about the featured
brands and ad films may be provided upon request).
The second step consists in coding the resulting segments with one or more
rhetorical figures and producing relevant descriptive statistics with the aid of
atlas.ti, as will be shown in the ensuing section.
The final step consists in applying multivariate mapping techniques and
factor analysis to the content analytic output with view to examining the overall
covariation patterns in the data-pool and interpreting the output.

5 Discussion of findings
The coding procedure resulted in 323 incidences of rhetorical figures (almost a 1:1
ratio between filmic segments and figures, even though this is not reflective of the
actual rhetorical structuration of the concerned films insofar as in many cases
each filmic segment was coded with more than one figure, while some segments
were not coded at all). Furthermore, the coding procedure with nuclear semes
(NSM) and classemes yielded a total of 492 semic incidences, of which 81% are
nuclear. Since the focus of this analysis rests with nuclear semes, the following
statistical output will revolve around laying out which nuclear semes shape the
cereals category’s sociolect and what differences emerge in the case of distinctive
brands” idiolects. Moreover, since, from a structuralist rhetorical point of view,
we are concerned primarily with relata, rather than with pro-filmic or ad expres
sive elements, as key sources for brand textual differentiation and furthermore as
sources for differential figurative advantages, the analysis will proceed by show
ing which rhetorical figures are responsible for shaping the cereals category
sociolect, followed by a focused analysis on the brands that make up our corpus.
Starting with the first part of our analysis, the semantic universe of the cereals
category was found to consist of the hierarchy of semes that is displayed in
Figure 1.
As per Figure 1, 75% of the cereals category’s semic universe is made up of
nine semes, namely, /energy/, /for the entire family/, /taste/, /for women/,
/wholegrain/, /superior performance/, /snappy, crackly, poppy sound/, /choco
latey/ and /shape/. Given the weight of this semic constellation we may claim that
these semes constitute the category’s sociolect. This certainly holds for semes,
such as /energy/, /wholegrain/, /superior performance/, but not for semes such
as /for women/, which is purely the province of Special K or /snappy, crackly,
poppy sound/ which is the province of Kellogg’s Rice Krispies. The two latter
semes emerged among the most popular ones that recur in the cereals’ sociolect
by virtue of their frequent recurrence in the concerned brands’ idiolects.
A more accurate depiction, at a preliminary level, of the centrality of each
seme in the cereals category’s sociolect may be yielded by attending to the aver
age number of brands where each of the nuclear semes occurs (Figure 2). In these
terms, seven nuclear semes recurred in more than 3 brands’ advertising dis
courses (from a diachronic point of view). More specifically, /taste/ emerged in 8
brand discourses, /wholegrain/ in 7, /energy/ in 6, /for the entire family/ in 5 and
the rest three semes, namely, /chocolatey/, /high in fiber/, and /superior perfor
mance/ in 3. The remaining semes emerged in 1–2 brand discourses and hence
it may be claimed that they constitute idiolectal aspects. Given that the six semes
/taste/, /wholegrain/, /energy/, /for the entire family/, /chocolatey/, and /superior

performance/ emerge more densely both in terms of number of brands as well as
frequency of recurrence, we may infer that they constitute the concerned cate
gory’s key value drivers or the semic drivers of the category’s sociolect.
In terms of associative type by brand, that is to what extent semes concerned
attributes, benefits or attitudes, as per Figure 3, we notice that attitude (39%) and
attribute (36%) related semes constitute the bulk of the category’s semic universe
in almost equal proportions, followed by benefits (24%).
As regards the differential distribution of types of semes by brand (Figure 3),
we notice that Kellogg’s breakfast cereals, Kellogg’s Special K, Kellogg’s All Bran,
Kellogg’s Crunchy Nut, Weetabix Minis, and Nestle Cheerios feature an above
category average incidence of attitudes; Kellogg’s Rice Krispies, Kellogg’s Coco
Pops, Kellogg’s Crunchy Nut, Kellogg’s Frosties, Weetabix Minis, Weetos, Nestle
Cheerios, Nestle Shreddies, and Nestle Shredded Wheat feature an above cate
gory average incidence of attributes; Kellogg’s Coco Pops, Kellogg’s All Bran,
Kellogg’s Frosties, Weetabix, Weetos, and Nestle Shreddies feature an above cate
gory average incidence of benefits.
Fig. 1: Cereals category hierarchy of semes (atlas.ti output)

Fig. 2: Distribution of semes by brand (atlas.ti output). Note: Cells highlighted in grey denote
that the seme is encountered at an above category average level in the concerned brand’s
semantic universe

Pursuant to the initial analyses pertaining to the category’s sociolect in terms
of semic drivers, further analyses were conducted with view to determining pat
terns of co occurrence among semes, modes of rhetorical configuration and brands.
To this end, a series of correspondence analyses, factor analysis, and multi
dimensional scaling analysis were conducted.
Fig. 3: Distribution of semes by brand according to associative type by attributes, benefits,
attitudes (atlas.ti output)
The first correspondence analysis was conducted among the data that make
up the brands x nuclear semes matrix (Figure 2), with view to discerning areas of
differential associations on a semic level, by taking into account the overall
covariance among the data. “Correspondence analysis is a method of data
analysis for representing tabular data graphically. Correspondence analysis is a
generalization of a simple graphical concept with which we are all familiar,
namely the scatterplot” (Greenacre 2007: 1). The
rationale of correspondence analysis consists of reducing a
dataset to as many salient dimensions as possible (Hoffman and
Franke 1986: 215). “The graphical relationships between the rows
and the columns of the table X that result from correspondence
analysis are based on the idea of representing all the row and column
categories and interpreting the relative positions of the points in terms of the
weights corresponding to the col umn and the row” (Hardle and Simar 2007:
306). “The proximity of a particular row to a particular column
indicates that this row (column) has a particularly important
weight in this column (row). In contrast to this, a row that is quite distant from
a particular column indicates that there are almost no observations in this
column for this row (and vice versa)” (Hardle and Simar 2007: 310). The
relative weight of association between the rows and columns of Figure 2 is
displayed in the graphical output of the respective
correspondence analysis (Figure 6), which was produced from
the contingency table’s (semes x brands) data with the program XLStat.
Since the probability value (pvalue) was found to be lower than

the Alpha level, the null hypothesis for independence between the variables
(semes, brands) was rejected (see Figure 4).
Further to the discernment of a significant association between brands and
semes, the cumulative percentage of the first two factors’ eigenvalues was calcu
lated with a view to discerning whether the significant association between
brands and semes was of sufficient weight that would allow for proceeding with
the graphical display of the correspondence analysis output. In these terms, as
per the results displayed in Figure 5, the cumulative eigenvalue of the first two
factors was 35.41% (F1 17.92%, F2 17.49%), suggesting a medium weight of associ
ation (in the context of a rule of thumb that suggests that a high associative weight
may be gauged from the incidence of an at least 60% cumulative percentage of
the first two factors; cf. Hardle and Simar 2007: 311).
This medium weight of association is evinced in the graphical output of Figure 6
where, with the exception of a strong association between Rice Krispies, All Bran,
Special K, Crunchy Nut, and Kellogg’s Breakfast Cereals and their corresponding
semic universes, the rest of the brands and semes tend to concentrate around the
middle of the map (cf. Greenacre 2007: 74), which is suggestive of an unclear and
tenuous link or that the link between semes and brands is diffuse. “The very low
inertia is seen in the closeness of the row profiles to the centroid” (Greenacre
2007: 79). In other words, the two dimensions of the correspondence analysis
Fig. 4: Chi-square independence test semes x brands (XLStat output)
Fig. 5: Eigenvalues and variance percentages for semes x brands (XLStat output)

map do not account sufficiently for the total variance in the data. “The asym
metric map functions well when total inertia is high, but it is problematic when
total inertia3 is small because the proﬁle points in principal coordinates are too
close to the origin for easy labeling” (Greenacre 2007: 80). However, “[correspon
dence analysis] should be regarded as a way of re-expressing the data in pictorial
form for ease of interpretation – with this objective any table of data is worth
looking at” (Greenacre 2007: 80). Insofar as the proximity of the data points on
the map allows for making inferences about the association between the examined
variables, the relative proximity of brands to semes yields an interesting picture
of the extent to which each a brand’s semic universe is sufficiently differentiated
from the rest.
3 “The total inertia of a cross-tabulation is a measure of how much variation there is in the table”
(Greenacre 2007: 81).
Fig. 6: Correspondence analysis scatterplot of rhetorical figures by brand

In order to further examine which semes and brands contributed differentially to
the total inertia of the table the individual row and column inertia scores for attri
butes and brands were summed and then percentages of each seme and brand
inertia scores were calculated, as per Figure 8. “The investigation of these compo
nents of inertia (analogous to an analysis of variance) plays an important sup
porting role in the interpretation of CA. They provide diagnostics which allow the
user to identify which points are the major contributors to a principal axis and to
gauge how well individual points are displayed” (Greenacre 2007: 81).
Fig. 7: Contribution of each seme and brand to the total inertia

The second correspondence analysis mapping exercise sought to determine
whether there is a significant association between brands and rhetorical figures
and if yes what is the weight of this association.
As per Figure 8, since the p value is lower than Alpha value the null hy
pothesis about the independence between the two variables was rejected and
hence it was confirmed that there is an association between the distribution of
the values of rhetorical figures by brand.
Further to the rejection of the null hypothesis about the independence of vari
ables, the weight of this association was determined by attending to the cumula
tive eigenvalue of the first two factors as per Figure 9.
The cumulative percentage (31.5%) of the first two factors (F1, F2) of Figure 9
is suggestive of a medium weight of association in the distribution of rhetorical
figures by brand. This is evinced in the scatterplot (Figure 10), where the data points
Fig. 8: Chi-square independence test rhetorical figures x brands (XLStat output)
Fig. 9: Eigenvalues and variance percentages for semes x brands (XLStat output)
From the calculation of individual row and column inertias we may discern the
following: In terms of rows (semes) /snappy, crackly, poppy sound/ and
/for women/ account for 22% of the entire rows’ inertia. The relatively high
contribution of these two attributes to the total variance may also be gauged
from their highly distanced placement from the centroid of the map (Figure 10).
In terms of columns, Special K and Rice Krispies account for 29.39% of the
entire columns’ inertia which may also be gauged from their highly distanced
placement from the centroid of the map (Figure 10).

from the two variables tend to be concentrated around the centroid of the map.
However, despite the medium weight of association we may still discern patterns
in the relationships between rhetorical figures and brands by cross-referring be
tween the indexed incidence of rhetorical figures by brand (Figure 12) and the
proximity of semes to figures in the correspondence analysis scatterplot (Figure 10).
More specifically, we notice a very specific pattern of differential modes of
rhetorical configuration in the cases of Rice Krispies (pareikonopoeia, onomato
poeia, and neologism), Weetabix, Weetabix Minis, and Nestle Shredded Wheat
(hyperbole), Shredded Wheat and Weetos (irony), Kellogg’s All Bran (parenthesis,
paronomasia), Nestle Cheerios (assonance, reshaption), pun, accolorance, rhyme
(Kellogg’s Special K and Coco-Pops).
Correspondence analysis is particularly pertinent in discerning differential
modes of rhetorical configuration among brand discourses in a given category, as
it takes into account the entire co-variance levels among the data of a brands x
figures matrix. The scatterplot output should always be checked against the orig
inal data in order to confirm whether the proximity between brands and figures
makes sense.
Fig. 10: Correspondence analysis scatterplot of rhetorical figures by brand

In order to further determine co-occurrence patterns (Oakes 1998; Biber et al.
2004; Gries 2009; Biber and Conrad 2009) among the nuclear semes that make up
the sociolectal semantic universe of the cereals product category and reduce
them to salient dimensions, a factor analysis was conducted. “In a factor analy
sis, the correlations among a large number of variables (i.e., the linguistic fea
tures) are identified, and the variables that are distributed in similar ways are
grouped together. Each group of variables is a factor – which is then interpreted
Fig. 11: Hierarchical ranking of the incidence of rhetorical figures across the entire corpus
(atlas.ti output; see Rossolatos 2013c, 2013d, 2013f for definitions of figures)

Fig. 12: Indexed incidence of rhetorical figures by brand. Note: Cells highlighted in yellow
denote that the rhetorical figure is encountered at an above category average incidence in the
concerned brand’s semantic universe

Fig. 13: Indexed incidence of rhetorical figures by seme. Note: Cells highlighted in yellow
denote that the rhetorical figure is encountered at an above category average incidence in the
concerned brand’s semantic universe

functionally as a dimension of variation” (Biber et al. 2004: 278). After nine itera
tions and pursuant to the recalculation of the factor loadings with varimax rota
tion4, a four-factor solution was yielded, as per Figure 14.
Positive factor loadings (i.e., with a value of above +0.3; as per the 0.3 cut-off
point suggested by Biber et al. 2004: 279) of individual semes by dimension (fac
tor) are highlighted in yellow (Figure 14). The labeling of each factor, as noted by
Biber et al. (2004) is a matter of interpretation, based on the function that each
underlying semantic dimension (i.e., factor) is intended to fulfill. “Because fea
tures with larger loadings are more representative of the factor, they are also more
useful in the functional interpretation of the factor)” (Biber et al. 2004: 279).
4 “The varimax rotation makes the interpretation easier by maximizing the variance of the
squared factors loadings by column. For a given factor, high loadings become higher, low load
ings become lower, and intermediate loadings become either lower or higher” (XLStat 2014).
Fig. 14: Factor analysis output (XLStat; after varimax rotation; 4-factor solution; factors with
loadings greater than 0.3 highlighted in yellow)

In this respect, by attending to the semantic contiguity of the positively loading
semes on each dimension, the following labels were attached to the four factors
(or semic drivers of the cereals category):
F1 safe and economic choice (positive loading of semes /uncertainty avoidance/,
/for the entire family/, /best ingredients/, /value-for-money/). This dimension is
clearly the territory of Kellogg’s cornflakes (cf. Figure 2).
F2 heritage and approval (positive loadings of semes /heritage/, /approval/,
/snappy, crackly, poppy sound/ and /stardom/). This dimension is clearly the ter
ritory of Kellogg’s Rice Krispies (cf. Figure 2).
F3 indulgence (positive loadings of semes /taste/, /inverted Britishness/, /ludic,
playful consumption experience and user profile/. This dimension is clearly the
territory of Kellogg’s Crunchy Nut (cf. Figure 2).
F4 health and wellness (positive loadings of semes /high in fiber/, /combating
bloatedness/, /for women/, /feeling good/). This dimension is mainly the terri
tory of Kellogg’s All Bran and Kellogg’s Special K (cf. Figure 2).
In order to discern the overall similarity in the semantic universe of the
brands that make up the selected corpus a multidimensional mapping exercise
was also conducted. The “multidimensional scaling technique [MDS] is a picto
rial representation of the relationships inherent in a dissimilarity matrix” (Oakes
1998: 253). “MDS enables us to map objects (brands) spatially, so that the rela
tive positions in the mapped space reflect the degree of perceived similarity be
tween the objects (the closer in space, the more similar the brands)” (Kohli and
Leuthesser 1993: 13).
Initially, a correlation matrix was produced with view to discerning brand
similarities across all nuclear semes. The correlation matrix is displayed in Figure
15.
Then MDS was applied to the correlation matrix, yielding two effective di
mensions, against which the brands were plotted, as per Figure 16.
As per the proximity of brands in the two-dimensional space of the resulting
MDS map we may discern that Special K and Kellogg’s All Bran share similar
semic universes, and the same holds for Kellogg’s Breakfast cereals, Weetabix
Minis, and Nestle Cheerios, for Kellogg’s Coco Pops and Weetos. Kellogg’s Crunchy
Nut, Weetabix, and Kellogg’s Rice Krispies maintain a quite differentiated semic
identity, based on their overall interaction patterns with the rest brands.
The MDS exercise offers a different outlook compared to the previously dis
played factor analytic exercise. Even though both factor analysis and MDS essen
tially reduce data to salient dimensions, MDS, by virtue of being customarily used
as a 2-dimensional solution (at most 3-D solution) affords to offer a snapshot of

Fig. 15: Correlation matrix of nuclear semes by nuclear semes for all brands in the corpus
(XLStat output)
Fig. 16: MDS map of all brands based on their degree of semic similarity (XLStat output)

brand similarities in terms of their semic universe in a 2-dimensional space. Fac
tor analysis offers a different outlook on the semic organization of the semantic
space of a product category, by yielding more salient dimensions against which
brands may be compared, and hence enlarging the discriminating scope of the
underlying semantic dimensions. The employment of both techniques in the ex
ploration of the modes of patterning of a product category’s sociolect affords to
yield complementary angles whereby a semic universe may be approached. Let it
be noted that whereas in traditional consumer research, MDS is applied to paired
evaluation test data (i.e., where consumers are requested to rate pairs of brands
based on their degree of perceived similarity), in corpus linguistics and, by exten
sion, in semiotic research, the degree of semic similarity among the brands that
make up a corpus is calculated indirectly, by applying multidimensional scaling
to the output of a correlation matrix. The corpus linguistic approach adopted in
the above exercise that consists of a projected brand image structure at an encod
ing stage may and should be compared and contrasted with emerging consumer
data in the context of tracking surveys and the production of perceptual maps (cf.
Myers 1996: 203).
6 Conclusions
This paper sought to demonstrate the usefulness of employing multivariate map
ping techniques that are amply employed in corpus linguistics and marketing
research in order to explore the distinctive modes of rhetorical-cum-semic config
uration of ad filmic texts. By using a series of techniques, such as correspondence
analysis, multidimensional scaling and factor analysis, different and salient facets
of brands’ rhetorical structures were explored in the context of total co-variation
patterns among the data that make up the selected brands’ semantic and rhetor
ical structure.
The above analyses are particularly pertinent for projecting brand image
structures at the very encoding stage of ad filmic texts, by attending to how a
brand’s semic structure alongside rhetorical relata emerge in specific ad filmic
segments. The resulting associations in consumers’ minds may then be compared
and contrasted through gap analyses with the above exploratory exercises in an
iterative and ongoing fashion in the context of tracking surveys.
The proposed methods of analysis for exploring distinctive modes of ad tex
tual configuration are instrumental for the attainment not only of a distinctive
semic structure, but, moreover, of differential figurative advantages qua distinc
tive modes of rhetorical configuration (Rossolatos 2013d). This is a novel facet in
the exploration of patterns of ad textual configuration that calls for further

inter-disciplinary research between corpus linguistics, semiotics, rhetoric, and
advertising.
References
Biber, Douglas, Susan Conrad Randi Reppen. 2004. Corpus linguistics: Investigating
language structure and use. Cambridge: Cambridge University Press.
Biber, Douglas Susan Conrad. 2009. Register, genre, and style. Cambridge: Cambridge
University Press.
Callister, Mark A. Lesa A. Stern. 2007. The role of visual hyperbole in advertising
effectiveness. Journal of Current Issues and Research in Advertising 29(2). 1–15.
De Chernatony, Leslie. 2006. From brand vision to brand evaluation: The strategic process of
growing and strengthening a brand. Oxford: Butterworth-Heinemann.
Dobni, Dawn George M. Zinkhan. 1990. In search of brand image: A foundation analysis.
Advances in Consumer Research 17. 110–119.
Durand, Jacques. 1970. Rhétorique et image publicitaire. Communications 15. 70–95.
Durand, Jacques. 1987. Rhetorical figures in the advertising image. In Jean Umiker-Sebeok (ed.),
Marketing and semiotics: New directions in the study of signs for sale, 295–318, Berlin:
Mouton de Gruyter.
Forceville, Charles J. 2007. Multimodal metaphor in ten Dutch TV commercials. Public Journal of
Semiotics 1(1). 15–34.
Forceville, Charles J. Eduardo Urios-Aparisi (eds.). 2009. Multimodal metaphor. Berlin:
Mouton.
Foss, Sonja. 2004. Framing the study of visual rhetoric: Toward a transformation of rhetorical
theory. In Charles A. Hill Marguerite Helmers (eds.), Defining visual rhetoric, 303–313,
New Jersey: LEA.
Foss, Sonja. 2005. Theory of visual rhetoric. In Sandra Moriarty, Kent Smith, Gretchen Barbatsis
Keith Kenney (eds.), Handbook of visual communication: Theory, methods and media,
141–152, New Jersey: LEA.
Goldenberg, Jacob David Mazursky. 2008. When deep structures surface: Design structures
that can repeatedly surprise. Journal of Advertising 37(4). 21–34.
Greenacre, Michael. 2007. Correspondence analysis in practice. London: Chapman and Hall.
Greimas, Algirdas J. 1966. Sémantique structurale. Paris: Presses Universitaires de France.
Greimas, Algirdas J. 1970. Du sens I. Paris: Seuil.
Greimas, Algirdas J. 1971. Narrative grammar: Units and levels. Modern Language Notes 86(6).
793–806.
Greimas, Algirdas J. 1976b. Maupassant: La sémiotique du texte. Paris: Seuil.
Greimas, Algirdas J. 1983. Du sens II. Paris: Seuil.
Greimas, Algirdas J. 1987. On meaning. Minneapolis: University of Minnesota Press.
Greimas, Algirdas J. Joseph Courtés. 1979. Dictionnaire raisonné de la théorie du langage.
Paris: Hachette.
Gries, Stefan. 2009. Quantitative corpus linguistics with R: A practical introduction. New York:
Routledge.
Groupe µ. 1970. Rhétorique générale. Paris: Larousse.
Groupe μ. 1992. Traité du signe visuel: Pour une rhétorique de l”image. Paris: Seuil.

Hardle, Wolfgang Leopold Simar. 2007. Applied multivariate statistical analysis. Berlin:
Springer.
Hatch, Mary J. James Rubin. 2006. The hermeneutics of branding. Journal of Brand
Management 14(1/2). 40–59.
Hoffman, Donna L. George R. Franke. 1986. Correspondence analysis: Graphical
representation of categorical data in marketing research. Journal of Marketing Research
23(3). 213–227.
Keller, Kevin L. 1998. Strategic brand management: Building, measuring and managing brand
equity. New Jersey: Prentice Hall.
Kohli, Chiranjeev S. Lance Leuthesser. 1993. Product positioning: A comparison of perceptual
mapping techniques. Journal of Product and Brand Management 2(3). 10–19.
Kostelnick, Charles David D. Roberts. 2010. Designing visual language: Strategies for
professional communicators. Boston, Massachusetts: Allyn and Bacon.
McQuarrie, Edward. 2008. A visit to the rhetorician’s workbench: Developing a toolkit for
differentiating advertising style. In Edward F. McQuarrie Barbara J. Phillips (eds.), Go
figure: New directions in advertising rhetoric, 257–276. New York: M. E. Sharpe.
McQuarrie, Edward F. David G. Mick. 1996. Figures of rhetoric in advertising language. Journal
of Consumer Research 22(4). 424–438.
McQuarrie, Edward Barbara J. Phillips. 2004. Beyond visual metaphor: A new typology of
visual rhetoric in advertising. Marketing Theory 4(1/2). 113–136.
Mick, David G. 1986. Consumer research and semiotics: Exploring the morphology of signs,
symbols and significance. Journal of Consumer Research 13(2). 196–213.
Mick, David G. 1987. Toward a semiotic of advertising story grammars. In Jean Umiker-Sebeok
(ed.), Marketing and semiotics: New directions in the study of signs for sale, 295–318,
Berlin: Mouton de Gruyter.
Mick, David G., James E. Burroughs, Patrick Hetzel Mary Yoko Brannen. 2004. Pursuing the
meaning of meaning in the commercial world: An international review of marketing and
consumer research founded on semiotics. Semiotica 152(1/4). 1–74.
Myers, James H. 1996. Segmentation and positioning for strategic marketing decisions.
Chicago: American Marketing Association.
Nöth, Winfried. 1990. Handbook of semiotics. Indianapolis: Indiana University Press.
Oakes, Michael P. 1998. Statistics for corpus linguistics. Edinburgh: Edinburgh University Press.
Rossolatos, George. 2012. Applying structuralist semiotics to brand image research. Public
Journal of Semiotics 4(1). 25–82.
Rossolatos, George. 2013a. Rhetorical transformations in multimodal advertising texts: From
general to local degree zero. Hermes 50. 97–118.
Rossolatos, George. 2013b. The relevance of structuralist rhetorical semiotics for ad textual
signification. Paper presented at the 5th
Conference on Rhetoric and Narratives in
Management, Barcelona, Spain, 25–27 March.
Rossolatos, George. 2013c. An anatomy of the multimodal rhetorical landscape of the world’s
most valuable brands. International Journal of Marketing Semiotics 1. 73–125.
Rossolatos, George. 2013d. //rhetor.dixit//: Understanding ad texts’ rhetorical structure for
differential figurative advantage. http://www.mediafire.com/download/
5zme42zb78p4xm7/rossolatos_rhetordixit_+ebook+version.zip (accessed 18 February
2014).
Rossolatos, George. 2013e. Brand equity planning with structuralist operations and operations
of rhetorical transformation. Kodikas 35(1). 153–168.

Rossolatos, George. 2013f. A methodological framework for conducting multimodal rhetorical
analyses of advertising films with ATLAS.ti. In Susanne Friese Thomas Ringmayr (eds.),
Atlas.ti user conference 2013: Fostering dialog on qualitative methods. Berlin: Berlin
Technical University Press.
Rossolatos, George. 2014. Brand equity planning with structuralist rhetorical semiotics. Kassel:
Kassel University Press.
Rossolatos, George. forthcoming. A methodological framework for projecting brand equity:
Putting back the imaginary into brand knowledge structures.
Sloane, Thomas (ed.). 2001. Encyclopedia of rhetoric. California: University of California Press.
Solomon, Michael Lawrence Greenberg. 1993. Setting the stage: Collective selection in the
stylistic context of commercials. Journal of Advertising 22(1). 11–23.
Stern, Barbara. 2001. Marketing images: Construct definition, measurement issues and theory
development. Marketing Theory 1(2). 201–224.
Tom, Gail Anmarie Eves. 1999. The use of rhetorical devices in advertising. Journal of
Advertising Research Jul/Aug. 39–43.
Wichmann, Anne. 2008. Speech corpora and spoken corpora. In Anke Ludeling and Merja Kyto
(eds.), Corpus linguistics: An international handbook vol. 1, 187–206, Berlin: Mouton.
Willerton, Russell. 2005. Visual metonymy and synecdoche: Rhetoric for stage-setting images.
Journal of Technical Writing and Communication 35(1). 3–31.
XLStat. 2014. Running a factor analysis with XLSTAT. http://www.xlstat.com/en/learning-
center/tutorials/running-a-factor-analysis-with-xlstat.html (accessed 18 February 2014).
Bionote
George Rossolatos (b. 1975) is an academic researcher and marketing practitioner
〈georgerossolatos123@gmail.com〉. His research interests include rhetoric, phe
nomenology, deconstruction, psychoanalysis, and cultural studies. His publica
tions include “//rhetor.dixit//: Understanding ad texts’ rhetorical structure for
differential figurative advantage” (2013); and “Brand equity planning with struc
turalist rhetorical semiotics” (2012, 2014).

Exploring the rhetorical semiotic brand image structure of ad films with multivariate mapping techniques

Recommended

Recommended

More Related Content

Similar to Exploring the rhetorical semiotic brand image structure of ad films with multivariate mapping techniques

Similar to Exploring the rhetorical semiotic brand image structure of ad films with multivariate mapping techniques (20)

More from //disruptiVesemiOtics//

More from //disruptiVesemiOtics// (20)

Recently uploaded

Recently uploaded (20)

Exploring the rhetorical semiotic brand image structure of ad films with multivariate mapping techniques